Acoustic Model Adaptation from Raw Waveforms with Sincnet (2019-09-30T00:00:00.000000Z)

TL;DR

It is shown that the parameterisation of the SincNet layer is well suited for adaptation in practice: it can efficiently adapt with a very small number of parameters, producing error rates comparable to techniques using orders of magnitude more parameters.

Abstract

Raw waveform acoustic modelling has recently gained interest due to neural networks' ability to learn feature extraction, and the potential for finding better representations for a given scenario than hand-crafted features. SincNet has been proposed to reduce the number of parameters required in raw-waveform modelling, by restricting the filter functions, rather than having to learn every tap of each filter. We study the adaptation of the SincNet filter parameters from adults' to children's speech, and show that the parameterisation of the SincNet layer is well suited for adaptation in practice: we can efficiently adapt with a very small number of parameters, producing error rates comparable to techniques using orders of magnitude more parameters.

Authors

S. Renals

4 papers

Ondrej Klejch

2 papers

P. Bell

3 papers

TL;DR

Abstract

Authors

References41 items

On Learning Interpretable CNNs with Parametric Modulated Kernel-Based Filters

Improving Children Speech Recognition through Feature Learning from Raw Speech Signal

Rapid Speaker Adaptation of Neural Network Based Filterbank Layer for Automatic Speech Recognition

Digital signal processing

Multi-timescale Feature-extraction Architecture of Deep Neural Networks for Acoustic Model Training from Raw Speech Signal

Acoustic Modeling from Frequency Domain Representations of Speech

Domain Adaptation Using Factorized Hidden Layer for Robust Automatic Speech Recognition

Learning to adapt: a meta-learning approach for speaker adaptation

Speaker Recognition from Raw Waveform with SincNet

The Microsoft 2017 Conversational Speech Recognition System

Factorised Representations for Neural Network Adaptation to Diverse Acoustic Environments

A deep neural network integrated with filterbank learning for speech recognition

Novel Unsupervised Auditory Filterbank Learning Using Convolutional RBM for Speech Recognition

Improving Children's Speech Recognition Through Out-of-Domain Data Augmentation

On combining i-vectors and discriminative adaptation methods for unsupervised speaker normalization in DNN acoustic models

Learning Hidden Unit Contributions for Unsupervised Acoustic Model Adaptation

Speech acoustic modeling from raw multichannel waveforms

Adam: A Method for Stochastic Optimization

Learning hidden unit contributions for unsupervised speaker adaptation of neural network acoustic models

Speaker adaptation of neural network acoustic models using i-vectors

Learning filter banks within a deep neural network framework

Understanding how Deep Belief Networks perform acoustic modelling

Learning a better representation of speech soundwaves using restricted boltzmann machines

Unleashing the killer corpus: experiences in creating the multi-everything AMI Meeting Corpus

Multi-Parameter Frequency Warping for Vtln by Gradient Search

The Fisher Corpus: a Resource for the Next Generations of Speech-to-Text

Robust recognition of children's speech

SRILM - an extensible language modeling toolkit

Maximum likelihood linear transformations for HMM-based speech recognition

Speaker normalization using efficient frequency warping procedures

Analysis of CNN-based speech recognition system using raw speech as input

Audio augmentation for speech recognition

Learning the speech front-end with raw waveform CLDNNs

“Keras

Acoustic modeling with deep neural networks using raw time signal for LVCSR

banks within a deep

The Kaldi Speech Recognition Toolkit

, Diego Giuliani , Matteo Gerosa , Christian Hacker , Martin Russell , Stefan Steidl , and Michael Wong , “ The PF STAR childrens speech corpus

Speech communication : human and machine

This Paper Is Included in the Proceedings of the 12th Usenix Symposium on Operating Systems Design and Implementation (osdi '16). Tensorflow: a System for Large-scale Machine Learning Tensorflow: a System for Large-scale Machine Learning

“Learning ﬁlter

Field of Study

Journal Information

Name

Page

Venue Information

Name

Type

URL

Alternate Names